Simple utility class for parsing and vectorizing vowpal wabbit data in python.

Vowpal wabbit format has a few very nice properties:
  * handles text and non-text very naturally
  * groups features by namespace
  * stores meta data, like importance weight and tags

This class tries to maintain fidelity for these properties and in addition:
  * allows for easy slicing and dicing of the data by leveraging pandas
  * supports direct extraction of large, sparse feature matrices grouped by namespace 
  * support simple feature pruning
  * pass-through processing of text using scipy text analyzers

To see it in action, type "python example.py".